How to Measure AEO: Track the Prompts That Send You Buyers

If you want to know how to measure AEO, track a fixed set of the prompts your ideal customer asks AI, then trace your real leads back to those prompts. Do that and you have receipts tied to buyers, and those receipts tell you more than any visibility score can on its own.

I keep seeing founders, including well-funded ones, say the same thing: the AEO advice contradicts itself, the agency dashboards tell different stories, and nobody can prove which AI answers sent them a real buyer. My lane is Human First measurement that ties AI answers to real leads. So here is how I think about it. This is me paying it forward, the same way I did when I shared how I optimized my website for AI search. This is what I do, and you can too.

What Is AEO And Why Is It So Hard To Measure?

AEO stands for answer engine optimization. In plain language, it means showing up when people ask ChatGPT, Claude, Perplexity, or Google's AI for an answer.

Here's the thing. Search used to give you a clean trail. Someone typed a keyword, saw a list of links, clicked yours, and your analytics could record the visit and where it came from. AI answers work differently. A buyer asks a question, gets a full answer in a chat window, and might read your name without ever clicking. If you want the full picture of how that works on one platform, I broke down ChatGPT search in a separate post.

Three things make this tricky to measure:

1. Answers change run to run. Ask the same question twice and you can get two different answers with different sources.
2. The trail is thin. Some AI tools pass referral data to your site. Many conversations end in the chat window.
3. The advice contradicts itself. Experts disagree, and dashboards can tell different stories.

So learning how to measure AEO comes down to one simple question: which AI answers sent you a real buyer? Everything in this post points back to that question.

How Do You Build An ICP Prompt Inventory?

Your ICP is your ideal customer profile, the exact person you want to work with. Your prompt inventory is the list of questions that person types into an AI before they know your name. Write those questions in their words, the way they would say them out loud, with the same frustration and the same plain language.

Start by splitting your prompts into two buckets:

• Buying-intent prompts. These are the questions someone asks when they are ready to hire, buy, or book. Think "who can help me set up AI for my coaching business" or "best AI coach for small business owners."
• Research prompts. These are the questions someone asks while they are still learning. Think "how do I use AI to save time in my business."

Both matter. Buying-intent prompts sit closest to a purchase. Research prompts show you what your ideal customer wants to understand first.

Next, keep this list as a fixed set in a living document you own. The whole point is to check the same prompts over time. If you change the list every week, you lose the ability to compare. Add new prompts when you learn something, and keep the core set steady.

Then check on a steady rhythm. I like a small core set weekly and the full set monthly. Because AI answers vary from run to run, one check is noisy. A steady rhythm gives you far more context than any single answer.

Here are my receipts. My AEO keyphrase work lives in a Notion database I own, where every post has a target question, a status, and a story angle. If you want to set up something similar, start with my guide on using Notion with AI for business.

How Do You Trace AI-Referred Leads Back To The Prompt?

This is where measurement gets real. Start from buyers and work backward.

Here is the simple process:

1. Add a "how did you hear about us" field to every intake form, booking page, and contact form. Make it an open text box so people can tell you in their own words.
2. Ask on the first call. When someone says "ChatGPT recommended you," follow up with "What did you ask it?" If they remember, write down their exact words.
3. Check your referral sources in analytics. Many AI tools show up as referral sources in your website analytics. Look for them and note which pages they landed on.
4. Match each lead back to your prompt inventory. If a buyer asked something close to a prompt on your list, tag it. If they asked something new, add it to the list.

Now some of your prompts have receipts. You can point to a real person who asked a real question, saw your name in an AI answer, and became a lead. That is a lead receipt you can stand on.

My weekly scorecard tracks answer-engine referral sessions as one of its sources. That gives me a running record of the answer-engine visits my analytics can see.

Make it a recurring process. Once the form field, the first-call question, and the weekly check are in place, you have one of those business loops that run without you. The loop collects the data, and you show up to read it and make decisions.

How To Measure AEO With Data You Own

Ownership is the foundation. Use a simple sheet or Notion database that you control. Every inbound lead gets two tags: the source and the prompt.

A simple setup looks like this:

• Lead name and date
• Source (ChatGPT, Claude, Perplexity, Google's AI, referral, social, other)
• Prompt (the exact question they asked, matched to your inventory when possible)
• Prompt type (buying intent or research)
• Outcome (call booked, proposal sent, became a client)

These fields give you a starting point for how to measure AEO with records you own. Over time you see which prompts bring buyers, which bring researchers, and which bring nothing yet.

When you own the data, it stays with you. If a vendor goes away, changes pricing, or changes how they calculate a score, your history stays put. Agency theater is a report you rent. Owned measurement is a receipt you keep.

One rule I hold myself to: honest baselines. When I added tracking, I recorded no data before the tracking day and never a fake zero. If you started tracking in March, your history starts in March. Filling in zeros for earlier months makes it look like you grew from nothing, and that story would be made up. An honest baseline shows exactly where your tracking begins.

This connects directly to measuring the ROI of AI in your business. Same principle. Tie the work to outcomes you can see, and keep the records yourself. If you want all of this to live in one place with the rest of your business knowledge, my guide on building a second brain with AI shows how I set mine up.

Why Do AEO Dashboards Disagree?

If you have ever looked at two AEO dashboards and seen two different stories, you are in good company. There are clear reasons for it.

• Different tools sample different prompts. One tool checks a set of questions it chose. Another checks a different set. They are measuring different things.
• They run on different days. Since AI answers shift from run to run, a check on Monday and a check on Thursday can look very different.
• They define "visibility" differently. One tool might count any mention. Another might count only a cited link. Check each tool's definition before you compare scores.

Visibility scores and share-of-voice charts are useful inputs. They can show you trends and spot gaps. Revenue is the scoreboard.

Here's the thing. Once you have your own prompt inventory and your own lead receipts, the disagreement stops being stressful. Every dashboard becomes one more input you check against your receipts. If a tool says you are winning on a prompt and your leads confirm it, great. If a tool says you are winning and no buyers ever mention that question, you know where to look next. Your receipts become the tiebreaker.

That is what leading from the front looks like with AEO measurement. You set the standard for your own business, and the tools serve that standard.

Your How To Measure AEO Action Plan

1. Write your ICP prompt inventory. List the questions your ideal customer asks AI before they know your name, in their words, split into buying intent and research.
2. Set your checking rhythm. Check a small core set of prompts weekly and the full set monthly, and log what you see.
3. Add the "how did you hear about us" question everywhere. Put it on every form and ask it on every first call, including "What did you ask the AI?"
4. Build your owned lead tracker. Create a sheet or Notion database that tags every inbound lead by source and by prompt, starting with an honest baseline from today.
5. Check dashboards against your receipts. Treat every score and chart as an input, and let real buyers and revenue decide what is working.

The Bottom Line

Learning how to measure AEO comes down to three moves: know the prompts your buyers ask, trace real leads back to those prompts, and keep the data yourself. Do that and the contradicting advice and the disagreeing dashboards get a lot quieter, because you have receipts. Start with one living document of prompts and one question on your intake form this week. When buyers share the questions they asked, you keep those receipts in a place you own.